别叫它战略，叫它行动议程：鲁梅尔特谈好战略的内核

Rumelt defines strategy as a design for overcoming a high-stakes challenge, a mix of policy and action. He traces the word to the Greek strategoi, the ten leaders elected in Athens to deal with the issues of the day, which could be invasion, plague or a temple fund. A challenge can be negative or positive, such as discovering oil in your back forty.

After years of teaching analytical tools such as matrices and the five forces, he came to see them as aids for analyzing a problem, not strategy itself. His wife Kate pushed him to define the term, and he concluded that all strategy is problem solving. That leads to a kernel of three elements: a diagnosis of the situation, a guiding policy for dealing with it, and coherent actions.

Diagnosis means deciding what to pay attention to and forming hypotheses about how things connect. The guiding policy is the core, and he contrasts it with a client's list of 17 priorities, noting that "priority" means the first thing, not everything that might matter. Coherent action means doing things that deal with the problem and do not fight each other, such as claiming you will grow faster while also raising profit margins and return on equity.

As a quick example of good strategy, Rumelt describes Microsoft reacting to AI: diagnose the challenge, pick a guiding policy of investing in a leader and building it into search, and then act. The difficulty is that many companies skip this. Some announce a robotics-and-AI future and then simply acquire a few companies, which he calls strategic assembly without any synthesis.

He says the standard bad strategy is a set of abstract profit or performance goals, which are really ambitions. He recalls helping a part of the US Department of Defense with a strategy for 17 intelligence agencies, where the draft essentially said they should coordinate better without asking why coordination was hard. A client who said the diagnosis was "not growing fast enough" was stating a wish, not explaining what held the company back.

Bad strategy also includes fluff, or "word salad", and incoherent actions, but he considers the lack of diagnosis and the mistaking of goals for strategy the two main causes. He gives the example of US education, where noting falling test scores is accurate but jumping to a policy of sending more people to college skips the question of why. In organizations as in politics, people argue over diagnoses, and you must resolve the argument before you can act.

Lenny summarizes that a missing diagnosis or missing actions signals a weak strategy, and asks whether there is a magic number. Rumelt says only that it should be a few, not 17. We work best when we concentrate, he says, and focus is a fundamental source of power in strategy.

He illustrates with a childhood story of a magnifying glass at summer camp, which would not burn a piece of cloth until a counselor offered a black thread. You need a source of power, a focus, and a target that can actually be affected. He uses US foreign policy as a case of diffused effort across many countries, and notes that complex organizations struggle to focus because different units serve different interests with different funding.

In a competition between equals the outcome is 50/50, so Rumelt says a strategy needs to exploit an asymmetry: being first to recognize something, a reputation, relationships or an invention. A reputation can even work against you if people do not expect you to do something new.

His example is IBM under Gerstner, who recognized that the company's respected status and access to every large corporation was its source of power and decided to "bear hug" its customers. Later, as computing moved to the cloud, those same large customers were the most hesitant, and IBM was disadvantaged by what it had inherited.

To find your own power, he asks how your company or team differs from others, what you know or possess that others do not, and sometimes redefines your space smaller until you can see it. He adds that business is exciting because it is unstable, with a constant changing of the guard, unlike government, which he says gets stultified.

Lenny reads out the powers listed in Rumelt's book, and Rumelt explains that the power new business models exploit today is the user base: network effects, where more users make a product more useful. He traces the idea to telephone systems, mentions Bill Gates and early BASIC, and then social media, Amazon and two-sided markets such as credit cards.

On generative AI, he says the suspicion is that size matters, because more data could make the AI better, just as Google's search data makes it hard for Bing to catch up. The way around a leader is to specialize and own an approach that is not the market leader's, though our attention is often drawn to giants. Venture capital funds startups racing to build network effects quickly.

Lenny cites Twitter as the ultimate network-effect example, since the product kept running after deep layoffs, a rename and algorithm changes. Rumelt wonders how it survives an 80 percent layoff and says he finds it much more interesting than five years ago.

Power is the reason the strategy makes sense, Rumelt says: some resource, skill or knowledge competitors cannot easily match, so the bet is not even. If someone asks whether to open a restaurant, he says the statistics suggest not to, so you need to explain where your asymmetry is, such as having trained under a successful chef.

Strategy is surprising when it works because we expect organizations to bumble. He cites the US in Afghanistan, where aims such as catching Osama bin Laden, nation-building, women's education and ending opium production pulled in different directions, and he argues that you need a focus on something achievable.

Since there is no science of strategy like the engineering of beams, he says much of it rests on analogy with previous human experience. That is why he loves business history, biographies and war history, and why he thinks it is hard to think strategically without access to other times and places.

Rumelt says the best history is written close to events, and he recommends reading old newspapers day by day to see how the 1929 crash unfolded as a surprise to everyone. Economists still disagree on why the Great Depression happened, and he finds it useful to put yourself back at that moment and form your own view.

He compares diagnosis to falling off a cruise ship and grabbing a floating piece of wood: relief at the first workable explanation is natural, but the most important intellectual tool is to think again and ask whether a bigger piece of wood is nearby. His own reading of 1929 was that demand for capital goods such as cars peaked once everyone who could afford one had one, a view he has not seen elsewhere.

He criticizes business schools for teaching theories because it is easier than teaching intellectual structure, and recalls a UCLA colleague whose MBA students complained to the dean about being assigned a book.

Asked how a PM should lay out a strategy, Rumelt says that after The Crux he would change a few things. First, diagnosis is not merely understanding the world but understanding the challenge you face, and the question to ask is what makes it hard. A CEO who wants to open in Australia must say why that is difficult, and that is where the strategy begins.

His two pieces of advice are to state the problem and to avoid the word strategy, using action agenda instead. He says that mission, vision and values are a different literature, and that you do not need them before you can have a strategy. Using his own ambitions at 25 as an example, he suggests listing all your ambitions, then asking which you can make progress on today and what is holding you back.

The challenge you choose must be important, because it is close to your ambition, and addressable, because you can actually do something about it. The result is a set of actions, not goals, spanning six months to a year in a small company or two to three years in a big one.

Rumelt explains that the crux is the hardest move in a climb, and that in business it is the hardest part of the problem. A climber who cannot handle the crux should not take on the climb. In design, skilled people focus on what makes a problem hard and see a way around it.

He gives several design stories. I. M. Pei wanted an entrance for the Louvre that would not obscure the building, which led to the glass pyramid. Elon Musk asked why rockets are hard to reuse. And on the Voyager design work he helped with, the question of locating Jupiter precisely was solved by taking a picture halfway there and triangulating against the stars.

Strategy, he says, is looking at the problem, asking what makes it hard and seeing a way to solve it, not choosing from a list of common strategies. That requires insight, which is scary because it is not guaranteed, but it comes from immersing yourself in the problem. Lenny's takeaway is to spend much more time diagnosing the biggest challenge, and Rumelt reiterates the action agenda framing.

On organizational politics, Rumelt says you need a hierarchy of power, because some mechanism must decide what to do, and a strategy cannot be "everything everybody wants." He compares it to a city committee that adds every suggestion, from park benches to birds. People inside organizations have private interests they may not state.

He recalls interviewing Donald Rumsfeld during the second Iraq invasion, when Rumsfeld said he had experts on everything but all of them disagreed and had their own agendas. Rumelt's view is that senior executives usually know most of what they need to know, but disagree on importance and have positions of power that would be hurt by some choices. Hence strategy inside organizations is a problem of diversity of interest and fear of action.

He thinks the leadership literature, focused on perfecting oneself, has displaced management and strategy, and that leaders hesitate to say "we're going to do this, and Bob is in charge of this." He cites Nokia, where he says engineers were replaced by lawyers and accountants and a matrix organization diffused power so that no one could act on building a touchscreen phone.

For startups, Rumelt says a founder is making a bet, like drilling for oil, and should be clear about the nature of the bet. His research on Silicon Valley startups found that the first target market and product generally do not work, and the ones that survive switch customers and products over a year or two, like a truffle hound searching. Founders must be of two minds: convinced they will win and willing to shift.

He notes that you cannot predict the future, using the early car industry as an example: the first cars sold in quantity in the US were electric delivery vehicles, and the First World War's truck production and returning mechanics helped gasoline take off. The action agenda for a startup is quick adaptation to changing conditions.

Lenny frames the founder's job as finding an important and achievable problem. Rumelt adds an exercise he called value denial: what should you be able to buy but cannot, such as reliably not losing luggage or hiring someone to remodel a house on time. He also describes imagining the perfect window, which should let in light and air but not bugs or noise, to show how asking how something ought to be reveals opportunities, though regulation can block some.

Rumelt's closing message is that strategy is about solving the most important problem you face that you can actually do something about, and you need focus on something doable and consistency. He is considering a fourth book, *Don't Do That*, inspired by his wife's ski lesson in Aspen, where an instructor cured her stance by yelling "don't do that" and nothing more.

In the lightning round he recommends reading biographies and histories rather than only strategy books, mentioning Clayton Christensen, Roger Martin's *Playing to Win*, Andy Grove's *Only the Paranoid Survive*, and Rockefeller's story, in which kerosene fell from a dollar to ten cents a gallon. His favorite interview question asks what hard thing the candidate did and why it was difficult, and about a strategy they find interesting, to test knowledge of the real world rather than theory.

He now earns a living mainly from public speaking and from "foundries": two-to-four-day sessions in which a CEO and seven or eight others list 25 to 50 problems, see they cannot do them all, and narrow to one or two important and addressable challenges. The outcome is an action agenda, not a mission and vision document. He invites listeners to share stories of creating strategies inside organizations.